Abstract:Classical multidimensional scaling only works well when the noisy distances observed in a high dimensional space can be faithfully represented by Euclidean distances in a low dimensional space. Advanced models such as Maximum Variance Unfolding (MVU) and Minimum Volume Embedding (MVE) use Semi-Definite Programming (SDP) to reconstruct such faithful representations. While those SDP models are capable of producing high quality configuration numerically, they suffer two major drawbacks. One is that there exist no theoretically guaranteed bounds on the quality of the configuration. The other is that they are slow in computation when the data points are beyond moderate size. In this paper, we propose a convex optimization model of Euclidean distance matrices. We establish a non-asymptotic error bound for the random graph model with sub-Gaussian noise, and prove that our model produces a matrix estimator of high accuracy when the order of the uniform sample size is roughly the degree of freedom of a low-rank matrix up to a logarithmic factor. Our results partially explain why MVU and MVE often work well. Moreover, the convex optimization model can be efficiently solved by a recently proposed 3-block alternating direction method of multipliers. Numerical experiments show that the model can produce configurations of high quality on large data points that the SDP approach would struggle to cope with.

Nonlinear Dimensionality Reduction with Relative Distance Comparison

Image Feature Optimization Based on Nonlinear Dimensionality Reduction

A Nonlinear Dimension Reduction Method with Both Distance and Neighborhood Preservation.

Dynamic Neighborhood Selection for Nonlinear Dimensionality Reduction

Nonlinear Dimensionality Reduction by Locally Linear Inlaying

Deep Nonlinear Sufficient Dimension Reduction

Stable Local Dimensionality Reduction Approaches

Nonlinear Dimensionality Reduction with Local Spline Embedding

Nonlinear Dimension Reduction Via Local Tangent Space Alignment

Locally Adaptive Nonlinear Dimensionality Reduction

Relative Distance-Based Laplacian Eigenmaps

Clustering-Based Nonlinear Dimensionality Reduction On Manifold

Robust nonlinear dimensionality reduction by topologically constrained semi-isometric embedding

Local Regularized Least-Square Dimensionality Reduction

Distortion-Free Nonlinear Dimensionality Reduction

A Tangent Distance Preserving Dimensionality Reduction Algorithm

Relative Manifold Based Semi-Supervised Dimensionality Reduction

Nonlinear Dimensionality Reduction Based on Statistical Criterions

C2DNDA: A Deep Framework for Nonlinear Dimensionality Reduction

Local Neighborhood Embedding for Unsupervised Nonlinear Dimension Reduction.

Convex optimization learning of faithful Euclidean distance representations in nonlinear dimensionality reduction